Re: [xacml] Erik absent from focus group this week

[Erik Rissanen <mirty@sics.se>]

Ron Williams wrote:

> I am troubled by the <LaterDelegateAttributeDesignator>. This may be
> abject ignorance or density on my part. However, does "Later" here
> seems to imply that it will be identified as a delegate later in
>  processing, but the text says "The previous */delegate/* is
> transformed into a */later delegate/* and the */issuer/* of Policy 2
> is the new */delegate/*" which implies that it should be called
> "PreviousDelegate" instead?


It is true that the LaterDelegate is really a previous delegate in the
processing model since the processing model does backwards chaining, but
it is a later delegate from the point of view of the policy writer. The
policy writer can use the LaterDelegate to constrain how delegation may
be done further down ("later") in the chain, beyond the immediate Delegate.

Personally I think the point of view of the policy writer is what we
should take. However, perhaps there is some other name that is more
appropriate? "TailDelegate"? "FurtherDelegate"? Any ideas?

Regards, Erik